iSharkFly-Open
/
python-peps
mirror of https://github.com/python/peps.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Checking in what we have so far. Still digesting some of Jim Jewett's 

inputs.
This commit is contained in:
Skip Montanaro 2004-03-23 16:41:17 +00:00
parent efebd8e76c
commit b2436faf3b
1 changed files with 206 additions and 168 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

370
pep-0318.txt
View File

@ -2,243 +2,281 @@ PEP: 318
Title: Function/Method Decorator Syntax
Version: $Revision$
Last-Modified: $Date$
Author: Kevin D. Smith <Kevin.Smith@theMorgue.org>
Author: Kevin D. Smith <Kevin.Smith@theMorgue.org>,
Jim Jewett <jimjjewett@users.sourceforge.net>,
Skip Montanaro <skip@pobox.com>
Status: Draft
Type: Standards Track
Content-Type: text/plain
Content-Type: text/x-rst
Created: 05-Jun-2003
Python-Version: 2.4
Post-History: 09-Jun-2003, 10-Jun-2003, 27-Feb-2004
Post-History: 09-Jun-2003, 10-Jun-2003, 27-Feb-2004, 23-Mar-2004
Abstract
========
The current method for declaring class and static methods
is awkward and can lead to code that is difficult to understand.
This PEP introduces possible new syntax which will place the
translation of instance methods to class/static methods at
the same point in the code as the method's declaration.
The current method for declaring class and static methods is awkward
and can lead to code that is difficult to understand. Ideally, these
transformations should be made at the same point in the code where the
declaration itself is made. This PEP introduces new syntax for
transformations of a declaration.
Motivation
==========
The current method of translating an instance method into a
class/static method places the actual translation at a different
point in the code than the declaration of the method. The
code below demonstrates this.
The current method of applying a transformation to a function or
method places the actual translation after the function body. For
large functions this separates a key component of the function's
behavior from the definition of the rest of the function's external
interface. For example::
def foo(self):
perform method operation
foo = classmethod(foo)
When the method is very short, it is easy to look ahead and see
that this is a class method. However, if the method is more than
15 lines or so, the translation into a class method is not
obvious. A solution to this problem is to move the translation
of the method to the same point as the method's declaration.
The proposed syntax, shown in the example below, is discussed
in the following sections.
This becomes less readable with longer methods. It also seems less
than pythonic to name the function three times for what is
conceptually a single declaration. A solution to this problem is to
move the transformation of the method closer to the method's own
declaration. While the new syntax is not yet final, the intent is to
replace::
def foo(self) as synchronized(lock), classmethod:
perform method operation
Proposal
Probably the simplest way to place the decorator that translates
an instance method to a class/static method is illustrated in the
code below.
def classmethod foo(self):
perform method operation
The code in this example will simply perform the following.
def foo(self):
perform method operation
def foo(cls):
pass
foo = synchronized(lock)(foo)
foo = classmethod(foo)
This syntax does not introduce any new keywords and is completely
backwards compatible with any existing code. The word between the
'def' and the actual name of the method is simply a reference to
a callable object that returns a new function reference.
This syntax could also be extended to allow multiple function
decorators in the form of a space delimited list as follows:
with an alternative that places the decoration in the function's
declaration::
def protected classmethod foo(self):
perform method operation
def foo(cls) using [synchronized(lock), classmethod]:
pass
which would be equivalent to the current form:
Background
==========
def foo(self):
perform method operation
foo = protected(classmethod(foo))
There is general agreement that syntactic support is desirable to the
current state of affairs. Guido mentioned `syntactic support for
decorators`_ in his DevDay keynote presentation at the `10th Python
Conference`_, though `he later said`_ it was only one of several
extensions he proposed there "semi-jokingly". `Michael Hudson raised
the topic`_ on ``python-dev`` shortly after the conference,
attributing the bracketed syntax to an earlier proposal on
``comp.lang.python`` by `Gareth
McCaughan`_.
While this syntax is simple and easy to read, it does become
cluttered and more obscure if you wish to allow arguments to be
sent to the function decorator.
.. _syntactic support for decorators: http://www.python.org/doc/essays/ppt/python10/py10keynote.pdf
.. _10th python conference: http://www.python.org/workshops/2002-02/
.. _michael hudson raised the topic: http://mail.python.org/pipermail/python-dev/2002-February/020005.html
.. _he later said: http://mail.python.org/pipermail/python-dev/2002-February/020017.html
.. _gareth mccaughan: http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=slrna40k88.2h9o.Gareth.McCaughan%40g.local
def synchronized(lock) classmethod foo(self):
perform method operation
Design Goals
============
Instead of placing the decorators in front of the function name,
a better place might be after it, as shown below. The word 'as' is
added simply as a separator to assist in readability.
The new syntax should
def foo(self) as synchronized(lock), classmethod:
perform method operation
* work for arbitrary wrappers, including user-defined callables and
the existing builtins ``classmethod()`` and ``staticmethod``
This syntax is quite clear and could probably be interpreted
by those not familiar with Python. The proposed syntax can be
generalized as follows:
* work with multiple wrappers per definition
'def' NAME '(' PARAMETERS ')' ['as' DECORATORS] ':'
* make it obvious what is happening; at the very least it should be
obvious that new users can safely ignore it when writing their own
code
where DECORATORS is a comma-separated list of expressions,
or a tuple. Using the latter form, the last example above
would look like:
* not make future extensions more difficult
def foo(self) as (synchronized(lock), classmethod):
perform method operation
* be easy to type; programs that use it are expected to use it very
frequently
This form make is possible for the list of decorators to
span multiple lines without using the line continuation operator.
* not make it more difficult to scan through code quickly. It should
still be easy to search for all definitions, a particular
definition, or the arguments that a function accepts
Alternate Syntaxes
* not needlessly complicate secondary support tools such as
language-sensitive editors and other "`toy parser tools out
there`_"
Other syntaxes have been proposed in comp.lang.python and
python-dev. Unfortunately, no one syntax has come out as a clear
winner in the lengthy discussions. The most common suggestions
are demonstrated below. The proposed syntax is also included
for easy comparison.
.. _toy parser tools out there: http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=mailman.1010809396.32158.python-list%40python.org
Proposed Syntax
Proposed Syntax
===============
def foo(self) as synchronized(lock), classmethod:
perform method operation
The currently proposed syntax is::
def foo(self) as (synchronized(lock), classmethod):
perform method operation
def func(arg1, arg2, ...) [dec1, dec2, ...]:
pass
Prefix Forms
The decorators are near the declaration of the function's API but are
clearly secondary. The square brackets make it possible to fairly
easily break long lists of decorators across multiple lines.
def [synchronized(lock), classmethod] foo(self):
perform method operation
Alternate Proposals
===================
def synchronized(lock), classmethod foo(self):
perform method operation
A few other syntaxes have been proposed::
# Same as above, but only identifiers are allowed
sync = synchronized(lock)
def sync, classmethod foo(self):
perform method operation
def func(arg1, arg2, ...) as dec1, dec2, ...:
pass
# Java-like
sync = synchronized(lock)
def @sync @classmethod foo(self):
perform method operation
The absence of brackets makes it cumbersome to break long lists of
decorators across multiple lines. The keyword "as" doesn't have the
same meaning as its use in the ``import`` statement.
Postfix Forms
::
def foo(self) [synchronized(lock), classmethod]:
perform method operation
def [dec1, dec2, ...] func(arg1, arg2, ...):
pass
def foo(self) (synchronized(lock), classmethod):
perform method operation
This form has the disadvantage that the decorators become visually
higher priority than the function name and argument list.
def foo(self) {'pre': synchronized(lock), 'classmethod': True}:
perform method operation
::
I'm not as fond of the forms that use '[ ]' since code like
'foo()[a]' looks as if you are getting the item 'a' from 'foo()'.
Although, this isn't as much of an issue when using '[ ]' in
a prefix form. The Java-like syntax adds new syntax that is
very arbitrary and is almost Perl-ish. In addition, since the
order in which the decorators are applied may matter, the last,
dictionary-style, syntax must be eliminated.
def func [dec1, dec2, ...] (arg1, arg2, ...):
pass
Implementation Issues
Quixote's Page Template Language uses this form, but only supports a
single decorator chosen from a restricted set. For short lists it
works okay, but for long list it separates the argument list from the
function name.
In the following example there are two function decorators:
synchronized(lock) and classmethod.
::
def foo(self) as synchronized(lock), classmethod:
perform method operation
using:
dec1
dec2
...
def foo(arg1, arg2, ...):
pass
Since these all appear within the operation of the 'def'
itself, it makes sense that synchronized, lock, and
classmethod must exist at the time that the definition
is executed. In addition, each of these arguments will be
evaluated before being applied to the compiled function.
This means that arguments like synchronized(lock) must
return a descriptor that will be applied to foo. Therefore,
the code above translates to:
The function definition is not nested within the using: block making
it impossible to tell which objects following the block will be
decorated. Nesting the function definition within the using: block
suggests block structure that doesn't exist. The name ``foo`` would
actually exist at the same scope as the using: block. Finally, it
would require the introduction of a new keyword.
def foo(self):
perform method operation
foo = classmethod(<returned-descriptor>(foo))
Current Implementation
======================
In the example above, <returned-descriptor> refers to the
descriptor returned by evaluating synchronized(lock).
Michael Hudson has posted a `patch`_ at Starship, which implements the
proposed syntax and left-first application of decorators::
It could easily be argued that the descriptors should be applied
in reverse order to make the application of the descriptor look
more like the resultant code. I tend to prefer this form.
def func(arg1, arg2, ...) [dec1, dec2]:
pass
def foo(self):
perform method operation
foo = <returned-descriptor>(classmethod(foo))
is equivalent to::
In either case, the modified function is bound to the function
name when the 'def' statement is executed.
def func(arg1, arg2, ...):
pass
func = dec2(dec1(func))
Open Issues
though without the intermediate creation of a variable named ``func``.
It is not clear at the moment if it is even possible to have
multiple decorators for a function. If decorators are required
to take a function/method and return a descriptor, it might
not even be possible to wrap multiple decorators. This should
be explored since the best syntax for multiple decorators
may not be the same as the best syntax for a single decorator.
.. _patch: http://starship.python.net/crew/mwh/hacks/meth-syntax-sugar.diff
Current Implementations
Examples
========
I am not personally familiar enough with Python's source to
implement the proposed syntax; however, Michael Hudson has
implemented the "square-bracketed" syntax (see patch at
http://starship.python.net/crew/mwh/hacks/meth-syntax-sugar.diff).
It should be fairly simple for the Python development team
to translate this patch to the proposed syntax.
Much of the discussion on ``comp.lang.python`` and the ``python-dev``
mailing list focuses on the use of the ``staticmethod()`` and
``classmethod()`` builtins. This capability is much more powerful
than that. This section presents some examples of use.
1. Define a function to be executed at exit. Note that the function
isn't actually "wrapped" in the usual sense.
::
def onexit(f):
import atexit
atexit.register(f)
return f
def func() [onexit]:
...
2. Define a class with a singleton instance. Note that once the class
disappears enterprising programmers would have to be more creative
to create more instances. (From Shane Hathaway on ``python-dev``.)
::
def singleton(cls):
return cls()
class MyClass [singleton]:
...
3. Decorate a function with release information. (Based on an example
posted by Anders Munch on ``python-dev``.)
::
def release(**kwds):
def decorate(f):
for k in kwds:
setattr(f, k, kwds[k])
return f
return decorate
def classmethod(f) [release(versionadded="2.2",
author="Guido van Rossum")]:
...
4. Enforce function argument and return types.
::
def accepts(*types):
def check_accepts(f):
def new_f(*args, **kwds):
for (a, t) in zip(args, types):
assert isinstance(a, t), \
"arg %r does not match %s" % (a,t)
return f(*args, **kwds)
assert len(types) == f.func_code.co_argcount
return new_f
return check_accepts
def returns(rtype):
def check_returns(f):
def new_f(*args, **kwds):
result = f(*args, **kwds)
assert isinstance(result, rtype), \
"return value %r does not match %s" % (result,rtype)
return result
return new_f
return check_returns
def func(arg1, arg2) [accepts(int, (int,float)),
returns((int,float))]:
return arg1 * arg2
Of course, all these examples are possible today, though without the
syntactic support.
Possible Extensions
===================
The proposed syntax is general enough that it could be used
on class definitions as well as shown below.
The proposed syntax is general enough that it could be used on class
definitions as well::
class foo(object) as classmodifier:
class foo(object) [dec1, dec2, ...]:
class definition here
However, there are no obvious parallels for use with other
descriptors such as property().
Conclusion
The current method of translating an instance method to a class
or static method is awkward. A new syntax for applying function
decorators should be implemented (proposed syntax shown below).
def foo(self) as synchronized(lock), classmethod:
perform method operation
The proposed syntax is simple, powerful, easy to read, and
therefore preserves those qualities of the Python language.
Use would likely be much less than function decorators. The current
patch only implements function decorators.
Copyright
=========
This document has been placed in the public domain.
This document has been placed in the public domain.